Abstract: A method for etching a silicon wafer includes: a main etching process using a first mixed gas to perform a plasma etching process to etch the silicon wafer until a pattern on the silicon wafer reaches a specified aspect ratio, the first mixed gas being configured to etch silicon and react with silicon to form non-volatile reaction products; and an auxiliary etching process using a second mixed gas to perform another plasma etching process, the second mixed gas being configured to react with silicon to generate the non-volatile reaction products, and a formation rate of the non-volatile reaction products in the auxiliary etching process being greater than a formation rate of the non-volatile reaction products in the main etching process. The main etching process and the auxiliary etching process are alternately performed at least once until the pattern on the silicon wafer reaches a specified etching depth.

Abstract: Techniques for performing prefetching for a ranking service in a microservice architecture are provided. In one technique, in response to receiving a content request, an entity identifier of an entity associated with the content request is determined, a host of a second service that is different than the first service is determined. The first service sends the entity identifier to the host of the second service. The second service retrieves entity feature data that is associated with the entity identifier. The first service identifies a set of content delivery campaigns, identifies the host of the second service, and sends the identity of the set of content delivery campaigns to the host of the second service. The host of the second service determines a ranking of the set of content delivery campaigns, a subset thereof is selected, and data about each selected campaign is transmitted over a computer network.

Abstract: The present disclosure provides methods, computing devices, and storage media for generating a training corpus. The method includes: mining out pieces of data from user behavior logs associated with a target application, each piece of data including a first behavior log and a second behavior log, the first behavior log including a user speech and a corresponding speech recognition result, the second behavior log belonging to the same user as the first behavior log and time-dependent with the first behavior log; and determining the user speech and the corresponding speech recognition result in each piece of data as a positive feedback sample or a negative feedback sample, based on the first behavior log and the second behavior log.

Abstract: Techniques for performing prefetching for a ranking service in a microservice architecture are provided. In one technique, in response to receiving a content request, an entity identifier of an entity associated with the content request is determined, a host of a second service that is different than the first service is determined. The first service sends the entity identifier to the host of the second service. The second service retrieves entity feature data that is associated with the entity identifier. The first service identifies a set of content delivery campaigns, identifies the host of the second service, and sends the identity of the set of content delivery campaigns to the host of the second service. The host of the second service determines a ranking of the set of content delivery campaigns, a subset thereof is selected, and data about each selected campaign is transmitted over a computer network.

Abstract: The present disclosure provides methods, computing devices, and storage media for generating a training corpus. The method includes: mining out pieces of data from user behavior logs associated with a target application, each piece of data including a first behavior log and a second behavior log, the first behavior log including a user speech and a corresponding speech recognition result, the second behavior log belonging to the same user as the first behavior log and time-dependent with the first behavior log; and determining the user speech and the corresponding speech recognition result in each piece of data as a positive feedback sample or a negative feedback sample, based on the first behavior log and the second behavior log.

Abstract: The invention discloses a method for controlling the time sharing starting of electronic ballasts and a delayed-started electronic ballast. According to the method, the electronic ballasts are delayed-started after being energized, and delay time for the delayed starting of the electronic ballasts is a random number acquired based on the temperature of the ballasts. The delayed-started electronic ballast includes an electronic ballast body and a delay switch. After the adoption of the technical scheme of the invention, a delayer with the delay time set by virtue of random numbers corresponding to different environmental temperature is additionally arranged in the ballast, and multiple ballasts connected in parallel in a circuit can be started at different time points after different delay time under the control of the same control switch, which remarkably reduces current impact on a power grid and reduces a voltage drop condition.

Abstract: The invention provides a High Intensity Discharge (HID) lamp acoustic resonance suppression method. Whether acoustic resonance is produced or not is judged, and a frequency-modulated current frequency provided by a ballast when the acoustic resonance is produced to fulfill the aim of suppressing the acoustic resonance. The ballast includes an acoustic resonance elimination module. The invention has the beneficial effects that due to the additional arrangement of the acoustic resonance elimination module, acoustic resonance is maximally suppressed, and the optimal stability of a discharge arc is ensured. These characteristics enable the product to stably run for each type of lamp tube and different power.

Abstract: Embodiments of the invention provide a substrate etching method, which includes: a deposition operation for depositing a polymer on a side wall of a silicon groove, an etching operation for etching the side wall of the silicon groove, and repeating the deposition operation and the etching operation at least twice. In the process of completing all cycles of the etching operation, a chamber pressure of a reaction chamber is decreased from a preset highest pressure to a preset lowest pressure according to a preset rule. The substrate etching method, according to various embodiments of the invention, avoid the problem of damaging the side wall, thereby making the side wall smooth.

Abstract: The invention provides a High Intensity Discharge (HID) lamp acoustic resonance suppression method. Whether acoustic resonance is produced or not is judged, and a frequency-modulated current frequency provided by a ballast when the acoustic resonance is produced to fulfill the aim of suppressing the acoustic resonance. The ballast includes an acoustic resonance elimination module. The invention has the beneficial effects that due to the additional arrangement of the acoustic resonance elimination module, acoustic resonance is maximally suppressed, and the optimal stability of a discharge arc is ensured. These characteristics enable the product to stably run for each type of lamp tube and different power.

Abstract: The invention discloses a method for controlling the time sharing starting of electronic ballasts and a delayed-started electronic ballast. According to the method, the electronic ballasts are delayed-started after being energized, and delay time for the delayed starting of the electronic ballasts is a random number acquired based on the temperature of the ballasts. The delayed-started electronic ballast includes an electronic ballast body and a delay switch. After the adoption of the technical scheme of the invention, a delayer with the delay time set by virtue of random numbers corresponding to different environmental temperature is additionally arranged in the ballast, and multiple ballasts connected in parallel in a circuit can be started at different time points after different delay time under the control of the same control switch, which remarkably reduces current impact on a power grid and reduces a voltage drop condition.

Abstract: Embodiments of the invention provide a substrate etching method, which includes: a deposition operation for depositing a polymer on a side wall of a silicon groove, an etching operation for etching the side wall of the silicon groove, and repeating the deposition operation and the etching operation at least twice. In the process of completing all cycles of the etching operation, a chamber pressure of a reaction chamber is decreased from a preset highest pressure to a preset lowest pressure according to a preset rule. The substrate etching method, according to various embodiments of the invention, avoid the problem of damaging the side wall, thereby making the side wall smooth.